A method for estimating the internal temperature of a planar solid oxide fuel cell stack
The temperature observer is constructed through the dimensionality reduction Longberg comprehensive algorithm, and the temperature distribution of the SOFC stack is estimated using the peripheral parameters of the stack, which solves the problem of high temperature monitoring costs under high temperature operating conditions, and achieves fast and accurate temperature management and temperature gradient distribution Estimates of
Patent Information
- Application Number
- CN202310075615.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-02-07
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2043-02-07
AI Technical Summary
Under the high temperature operating conditions of solid oxide fuel cell (SOFC) stacks, it is impossible to install thermocouples and other sensors at low cost to directly obtain the internal temperature distribution of the stack, resulting in high temperature monitoring costs and difficult to achieve accurate temperature management.
Using the dimensionality reduction Longberg comprehensive algorithm, a temperature observer is constructed through limited and easy-to-measure parameters on the periphery of the stack. Using the molar flow rate of the air inlet, the molar flow rate of the fuel inlet and the outlet temperature information, the temperature distribution of the internal stack is quickly and accurately estimated. .
It reduces the cost of obtaining temperature information in the stack, provides reliable temperature management and control, ensures the safe and stable operation of the stack, and can estimate the temperature gradient distribution in real time.
Smart Images

Figure CN115966733B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fuel cells, and more particularly to a method for estimating the internal temperature of a planar solid oxide fuel cell stack. Background Art
[0002] At present, solid oxide fuel cell technology can directly convert the chemical energy in hydrogen, carbon monoxide and other hydrocarbons into electrical energy, avoiding the intermediate step of converting mechanical energy into electrical energy. Therefore, the power generation efficiency of solid oxide fuel cells is higher than that of most power generation devices, making it one of the most promising green energy power generation conversion technologies. SOFC can be widely used in various fields, including power sources for new energy vehicles, joint microgrids, distributed power stations and many other fields. Although SOFC technology has made major breakthroughs in many aspects, there are still many technical barriers to overcome in the commercialization process of SOFC. The most striking challenge is to monitor and control the temperature and temperature gradient inside the stack.
[0003] The optimal operating temperature of the SOFC stack is generally between 600 and 900 °C. Maintaining a suitable stack operating temperature is crucial to achieving high performance and long life of the SOFC system. On the one hand, if the operating temperature of the stack is lower than the optimal operating temperature, the materials inside the stack will not be able to reach the activation state, resulting in unsatisfactory electrical output performance of the stack. On the other hand, if the operating temperature of the stack is higher than the optimal temperature, it will exceed the thermal tolerance of the stack materials, accelerate the aging of the stack components and reduce the durability of the stack materials. More importantly, the PEN and metal connectors inside the stack are key components of the stack. They have different thermal expansion coefficients, so a certain temperature gradient will be generated inside the stack. A large temperature gradient will cause excessive thermal stress in the PEN and connector components, resulting in severe deformation or even rupture of the battery cells, and ultimately shortening the service life of the stack. Therefore, mastering the temperature distribution inside the stack is conducive to controlling the temperature and temperature gradient inside the stack, which is an important guarantee for the long-term stable and high-performance operation of SOFC.
[0004] However, due to the strict requirements on sealing in the high-temperature operating environment of the stack, it is impossible to install a temperature monitoring device such as a thermocouple inside the SOFC stack at low cost to directly obtain the temperature information inside the stack without damaging the stack material and reducing the stack performance. Therefore, it is a very effective solution to design a temperature observer to quickly and accurately estimate the temperature distribution inside the stack through limited and easy-to-measure stack peripheral parameters.
[0005] Regarding the technical research on the temperature of the fuel cell stack, the current research mainly focuses on the maximum operating temperature or average temperature control of the fuel cell stack. There is very little technical research on estimating the internal temperature distribution of the fuel cell stack through the finite parameters of the SOFC fuel cell stack periphery. Therefore, it is very important to propose a method for estimating the internal temperature distribution of the SOFC fuel cell stack for SOFC technology. The existing technology has high requirements for sealing under the high-temperature operating conditions of the SOFC fuel cell stack, and it is impossible to install sensors such as thermocouples inside the fuel cell stack at low cost to directly obtain the temperature distribution inside the fuel cell stack.
[0006] Therefore, how to reduce the cost of obtaining the temperature distribution inside the SOFC stack under the high sealing requirements of the high-temperature operation of the SOFC stack and accurately obtain the temperature distribution is an urgent problem to be solved by those skilled in the art. Summary of the invention
[0007] In view of this, the present invention provides a method for estimating the internal temperature of a planar solid oxide fuel cell stack, which can reduce the cost of monitoring the internal temperature of a solid oxide fuel cell stack and achieve rapid and accurate acquisition of the temperature distribution inside the stack.
[0008] In order to achieve the above object, the present invention adopts the following technical solution:
[0009] A method for estimating the internal temperature of a planar solid oxide fuel cell stack comprises the following steps:
[0010] Step 1: Construct a single cell mechanism model of the battery stack;
[0011] Step 2: Use the quasi-static assumption method to simplify the mechanism model and obtain a simplified model;
[0012] Step 3: Convert the simplified model into a state space model;
[0013] Step 4: Perform Taylor series expansion on the state space model to obtain a linearized state space model;
[0014] Step 5: Traverse the condition numbers of the observability matrix of all variable combinations in the linearized state space model to determine the observed variable combination; select the observable variable combination with the small condition number;
[0015] Step 6: Use the reduced-dimensional Romberg synthesis algorithm to construct a stack temperature observer based on the combination of observed variables, collect the stack peripheral measurement data and input it into the stack temperature observer to obtain the estimated state of the stack internal temperature distribution.
[0016] Preferably, the single cell sheets of the battery stack are divided into nodes according to the airflow direction, and mechanism modeling is performed according to the nodes. Each node mechanism model consists of a mass conservation model, an energy conservation model and an electrical characteristic model; the energy conservation model includes a solid layer temperature dynamic model and an air layer temperature dynamic model.
[0017] Preferably, a quasi-static assumption method is used to reduce the number of state variables in the node to obtain a simplified model; the simplified model includes two state variables.
[0018] Preferably, Taylor series expansion is performed at the steady-state operating point of the state-space model, and the obtained linearized state-space model is expressed as:
[0019]
[0020]
[0021] (x0,u0) is the steady-state operating point; x and u are the state variable and input variable of the stack system respectively; δx=x-x0, δu=u-u0, α(δx,δu) and β(δx,δu) are high-order small terms. After ignoring the high-order small terms, the state space equation of the stack nonlinear coefficient after linearization in the neighborhood of the steady-state operating point is obtained:
[0022]
[0023] δy=g(x,u)-g(x0,u0)=Cδx+Dδu
[0024] A, B, C, and D are all coefficient matrices of the linearized state-space model.
[0025] Preferably, the determined combination of observed variables includes input air flow, fuel gas flow and stack outlet terminal temperature.
[0026] Preferably, the process of constructing the stack temperature observer is:
[0027] Step 61: Select negative numbers as eigenvalues in the left half plane of the complex plane of the linearized state space model to form an eigenvalue group. Calculate the expected characteristic polynomial, the expression is:
[0028]
[0029] in, is the coefficient of the expected characteristic polynomial; s is the complex variable in control theory; n is the dimension of the coefficient matrix A in the linearized state space model; q is the rank of the coefficient matrix C in the linearized state space model; rankC = q;
[0030] The number of expected eigenvalue groups is determined according to the dimension n of the linearized state space model matrix A and the rank of the matrix C. If all the eigenvalue groups are located in the left half plane of the imaginary axis of the complex plane, the stability of the system can be guaranteed.
[0031] Step 62: The expected characteristic polynomials are formed into a (nq)×(nq) basic matrix F0, expressed as:
[0032]
[0033] Step 63: Select any non-single odd-number matrix in the basic matrix F0 Calculate the inverse R -1 , calculate the dimension matrix F = RF0R -1 ; Indicates limiting the matrix dimension to (nq)×(nq);
[0034] Step 64: Take any constant matrix in the basic matrix F0 Let the dimension of the {F,G} fully controllable matrix G be (nq)×q, and the expression is:
[0035]
[0036] Among them, F is the matrix to be solved; F n-q-1 It is the construction of the controllable matrix in linear system theory, which is used to determine whether {F, G} is completely controllable and whether the matrix G is correct. If it is not correct, it needs to be reselected.
[0037] Step 65: Solve the (nq)×n dimension matrix T of the Sylvester equation TA-FT=GC according to the coefficient matrix A and the coefficient matrix C, the matrix G, and the matrix F in the linearized state space model;
[0038] Step 66: If q>1, go to step 67; if q=1, construct a non-singular matrix based on the coefficient matrix C and matrix T in the linearized state space model And proceed to step 68;
[0039] Step 67: Determine if the matrix is non-singular If it is not singular, go to step 68; if it is singular, return to step 63;
[0040] Step 68: Construct the matrix H based on the matrix T and the coefficient matrix B in the linearized state space model, expressed as:
[0041] H = TB;
[0042] Step 69: Calculate the matrix Q based on the non-singular matrix P and perform block processing. The expression is:
[0043]
[0044] Q1 is an n×q matrix, Q2 is an n×q matrix;
[0045] Step 610: construct a reduced-dimensional Romberg stack temperature observer based on matrix F, matrix G, matrix H, matrix Q1, and matrix Q2, and the expression is:
[0046]
[0047]
[0048] Wherein, y represents the output of the battery stack system; u represents the input of the battery stack system; is the estimated state of the temperature distribution inside the battery stack; z is the intermediate state quantity of the estimated temperature inside the battery stack, represents the differential of z;
[0049] Step 611: Collect peripheral measurement data of the stack according to the observation variable combination, input the peripheral measurement data of the stack into the stack temperature observer, and obtain the estimated state of temperature distribution inside the stack. The input air flow rate, fuel gas flow rate and stack outlet terminal temperature in the observation variable combination correspond to the air inlet molar flow rate, fuel inlet molar flow rate and stack outlet temperature information in the stack peripheral measurement data, respectively.
[0050] It can be seen from the above technical scheme that, compared with the prior art, the present invention discloses a method for estimating the internal temperature of a flat-plate solid oxide fuel cell stack, and constructs a reduced-dimensional Romberg temperature observer through limited and easily measurable parameters around the SOFC stack. Only the air inlet molar flow rate, fuel inlet molar flow rate and temperature information of the stack outlet that are easily measured around the stack are needed to accurately estimate the temperature distribution of the stack, thereby reducing the cost of obtaining the internal temperature information of the stack, providing a reliable basis for the temperature management and control inside the stack, and solving the problems of high cost of internal temperature monitoring of the solid oxide fuel cell stack and difficulty in directly installing sensors for measurement. In addition, the temperature distribution inside the stack can be estimated quickly, accurately and in real time, thereby obtaining the distribution of the temperature gradient and ensuring the safe and stable operation of the stack. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0052] Figure 1 The accompanying drawing is a schematic flow chart of a method for estimating the internal temperature distribution of a SOFC stack provided by the present invention;
[0053] Figure 2The accompanying drawing is a schematic diagram of the structure and principle of a SOFC single cell provided by the present invention;
[0054] Figure 3 The accompanying drawing is a schematic diagram of the division of the flat plate type co-directional flow node provided by the present invention;
[0055] Figure 4 The accompanying drawing is a schematic diagram of the electrical characteristic curve of the SOFC cell provided by the present invention;
[0056] Figure 5 The accompanying drawing is a schematic diagram of voltage-based control provided by the present invention;
[0057] Figure 6 The accompanying drawing is a simplified structural schematic diagram of a SOFC stack and a temperature state observer provided by the present invention;
[0058] Figure 7 The accompanying drawing is a schematic diagram of the structure of the reduced-dimensional Romberg stack temperature state observer provided by the present invention;
[0059] Figure 8 The accompanying drawing is a schematic diagram of the comprehensive algorithm flow of the reduced-dimensional Romberg stack temperature state observer provided by the present invention. DETAILED DESCRIPTION
[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0061] The embodiment of the present invention discloses a method for estimating the internal temperature of a flat-plate solid oxide fuel cell stack. By designing a stack temperature observer, a reduced-dimensional Romberg synthesis algorithm is used to estimate the internal temperature distribution of the stack. The method mainly includes six parts: mechanism modeling, model simplification, state space model conversion, linearization processing, selection of observation variables and stack temperature observer, and includes the following steps:
[0062] S1: Mechanistic modeling;
[0063] First, the nodes of the single cell of the battery stack are divided according to the direction of airflow, and the mechanism model is built according to the nodes. Each node mechanism model consists of a mass conservation model, an energy conservation model and an electrical characteristic model. The mass conservation model mainly calculates the molar fractions of oxygen, nitrogen, hydrogen and water in the cathode and anode flow channels. The energy conservation model regards the metal connector, PEN and anode flow channel as solid layers, and establishes a solid layer temperature dynamic model. In the air flow channel, due to the fast gas flow rate and insufficient heat exchange, a separate air layer temperature dynamic model needs to be established. In the electrical characteristic model, the activation loss, ohmic loss and concentration loss are replaced by equivalent resistance, and based on the voltage control method, it is assumed that the voltage of each node is a constant value to quickly obtain the current size of a single cell node.
[0064] S2: Simplification of the model;
[0065] The established mechanism models have a high degree of coupling and high computational complexity. In order to reduce the complexity of the mechanism model and the number of state variables in the node, the quasi-static assumption is used to simplify the model. The quasi-static assumption assumes that the concentration of each reaction substance in the anode and cathode flow channels reaches a steady state in an instant, eliminating Four reaction concentration state variables are used to reduce the complexity of the model. are the mole fractions of hydrogen, water, oxygen, and nitrogen in the jth node (j=1, 2, 3, 4, 5), respectively;
[0066] S3: Transformation of state-space model;
[0067] The simplified model only has and Two state variables, and are the air layer temperature and solid layer temperature of the jth node respectively; the simplified model is converted into a state space model and abstracted into the following form:
[0068]
[0069] y=g(x,u)
[0070] Among them, x and u are the state variable and input variable of the battery stack system, respectively, specifically:
[0071]
[0072]
[0073] and and and and are the air layer temperature and solid layer temperature of nodes 1-5 respectively; represents the air inlet molar flow rate; represents the fuel inlet molar flow rate;
[0074] S4: linearization processing;
[0075] The obtained nonlinear state space model is subjected to Taylor series expansion at its steady-state operating point (x0, u0) to obtain the linearized state space model of the SOFC stack;
[0076] S5: Selection of observed variables;
[0077] The observed variables are the variable combinations that need to be input to the designed temperature observer. The selection of observed variables should first consider the observability of the system and secondly whether they are easy to measure in actual engineering. Therefore, the condition number of the observability matrix of all variable combinations is adopted, that is, the condition number of the observability matrix of the random combination of variables x, and the observed variable combination is determined to be the input air flow, fuel gas flow and temperature information at the outlet of the stack.
[0078] S6: stack temperature observer;
[0079] The stack temperature observer is the core of this technical solution. The observer uses the reduced-dimensional Romberg synthesis algorithm to estimate the temperature distribution inside the stack. By observing the variable combination of air flow, fuel inlet flow and stack outlet terminal temperature information, the stack internal temperature distribution can be estimated quickly and accurately. Indicates that, specifically Q1 and Q2 are matrices that need to be solved by the reduced-dimensional Romberg algorithm, and z is the intermediate state quantity of the internal temperature estimation of the battery stack. Its differential equation expression is: F, G, and H are matrices of corresponding dimensions to be solved, y is the system output of the battery stack, and u is the system input of the battery stack;
[0080] The specific steps to construct a reduced-dimensional Romberg stack temperature observer are as follows.
[0081] S61: Eigenvalue group Compute the expected characteristic polynomial:
[0082]
[0083] in, is the coefficient of the expected characteristic polynomial; s is the complex variable in control theory; n is the dimension of the coefficient matrix A in the linearized state space model; q is the rank of the coefficient matrix C in the linearized state space model; rankC = q;
[0084] S62: Form a (nq)×(nq) basic matrix F0:
[0085]
[0086] S63: Take any non-single odd-number matrix Calculate the inverse R -1 , calculate F = RF0R -1 ;
[0087] S64: Take any constant matrix Make {F,G} completely controllable, that is
[0088]
[0089] Among them, F is the matrix to be solved; F n-q-1 It is the construction of the controllable matrix in linear system theory, which is used to determine whether {F, G} is completely controllable and whether the matrix G is correct. If it is not correct, it needs to be reselected.
[0090] S65: solving the (nq)×n dimensional matrix T of the Sylvester equation TA-FT=GC according to the coefficient matrix A and the coefficient matrix C, the matrix G, and the matrix F in the linearized state space model;
[0091] S66: If q>1, proceed to the next step; if q=1, the coefficient matrix C and matrix T in the linearized state space model form a non-singular matrix in the next step And skip the next step and go directly to S68;
[0092] S67: Judgment If it is not singular, go to the next step; if it is singular, go to S63;
[0093] S68: Calculate H=TB;
[0094] S69: Calculate and block Q1 is an n×q dimensional matrix, Q2 is an n×q dimensional matrix;
[0095] S610: In summary, the main form of the reduced-dimensional Romberg state observer is:
[0096]
[0097]
[0098] in, is the estimated state of the temperature distribution inside the stack; z is the estimated intermediate state quantity of the temperature inside the stack, represents the differential of z;
[0099] S611: Input the collected peripheral measurement data of the fuel cell stack into the fuel cell stack temperature observer to obtain the estimated state of the internal temperature distribution of the fuel cell stack.
[0100] Example
[0101] The technical scheme for estimating the temperature distribution inside the SOFC stack is as follows: Figure 1 As shown, it mainly includes three parts: mechanism modeling, model simplification and linearization, and stack temperature observer.
[0102] 1. Mechanism modeling
[0103] The risk and cost of SOFC stack system experiments are high. In order to study the temperature distribution inside the SOFC stack, it is necessary to establish a mechanism model that can accurately describe the inherent physical properties of the stack. A single cell consists of PEN (cathode, electrolyte, anode), connector, air flow channel and fuel flow channel. The structure and principle diagram are shown in the figure. Figure 2 In order to more clearly analyze the distribution characteristics of various variables inside the battery stack, the finite element method is used to divide the battery stack into 5 nodes at equal distances according to the airflow direction. The mechanism model of each node consists of 3 sub-models: mass conservation, energy conservation and electrical characteristics. The node division of a single cell is as follows: Figure 3 shown.
[0104] Mass conservation submodel:
[0105] In the mass conservation submodel, the dynamic changes of physical quantities such as the molar flow rate, mole fraction, and gas pressure of each substance over time when the cathode and anode of each node undergo electrochemical reactions are mainly considered. The mole fraction of gas in the air and fuel flow channels is calculated in the mass conservation model, and the specific calculation method is as follows:
[0106]
[0107]
[0108]
[0109]
[0110] Where: are the mole fractions of hydrogen, water, oxygen and nitrogen in the jth node (j=1, 2, 3, 4, 5), respectively. and are the molar flow rates of air and fuel at the outlet of the jth node, I j is the current of the jth node, F1 is the Faraday constant, and are the total number of moles of gas in the jth node in the cathode and anode flow channels, respectively. The specific calculation method is as follows:
[0111]
[0112]
[0113]
[0114]
[0115]
[0116] In the formula, and are the volumes of the anode and cathode flow channels at the jth node, and are the temperatures of the air and solid layer at the jth node, and are the pressures of the anode and cathode of the jth node, respectively, and R is the universal gas constant. When j = 1, and equal and
[0117] Energy conservation submodel:
[0118] A single battery cell is composed of PEN, metal connector, anode and cathode channels. The heat transfer coefficient of each part is different, so the battery stack has different temperature dynamic response characteristics when working. Since the fuel flow rate is slow, there is sufficient heat exchange with PEN and metal connector. Therefore, the temperature of the metal connector, PEN layer and anode flow channel is unified as the solid layer temperature to establish a temperature dynamic model. The equivalent schematic diagram of the solid layer is shown in the figure. Figure 3 As shown. Since the air flow rate is much greater than the fuel flow rate, there is not enough time for the metal connector and PEN in contact with it to exchange heat, so the air layer needs to establish a separate temperature dynamic model. The 2 temperature layer model is as follows:
[0119] The air layer energy conservation submodel of the jth node is as follows:
[0120]
[0121] Where: and are the specific heat capacity and thermal enthalpy corresponding to O2 respectively; and are the specific heat capacity and thermal enthalpy of nitrogen respectively; when j = 1, and equal and k1 is a constant.
[0122] When j = 2, 3, 4, the solid layer energy conservation submodel is as follows:
[0123]
[0124] Where: pen ,C pen ,τ pen are the density, specific heat capacity and thickness of the PEN layer respectively; ρ ic ,C ic ,τ ic are the density, specific heat capacity and thickness of the metal connector respectively, and k2, k3, k4 and k5 are constant coefficients.
[0125] Since the first node (j=1) and the last node (j=5) of the solid layer only conduct heat with the adjacent nodes in direct contact, their mathematical model expressions are somewhat different from those of the intermediate nodes (j=2, 3, 4).
[0126] When j = 1 and j = 5, the solid layer energy conservation submodel is as follows:
[0127]
[0128]
[0129] Electrical characteristics model:
[0130] In the electrical characteristic sub-model, the main focus is on the relationship between current, voltage and power. According to the idea of dividing the nodes of a single cell according to the direction of airflow, the entire stack can be regarded as a cell connected in parallel between nodes, and a SOFC stack between cells in series. Therefore, the voltage of each node of a SOFC single cell is equal, which is equal to the open circuit voltage minus the three polarization losses. The specific expression is as follows:
[0131]
[0132] Schematic diagram of the electrical characteristics of SOFC cells Figure 4 As shown, V cell and V j are the operating voltage of a single cell and the operating voltage of node j respectively; are the ohmic loss, activation loss and concentration loss of the jth node, respectively, V loss is the total loss of node j, V ocv is the open circuit voltage at node j.
[0133] Due to the high complexity of calculating the three polarization losses, the equivalent resistance method is used to replace the three losses. The specific form is as follows:
[0134]
[0135] in represents the equivalent resistance of the jth node, due to the ohmic resistance Activation resistance and concentration resistance Average temperature Function of equivalent resistance Polynomial fitting can be performed as follows:
[0136]
[0137]
[0138] n is the equivalent resistance The highest degree of the polynomial, a0, a1, a2, ...a n is the equivalent resistance The coefficients of the polynomial.
[0139] The present invention is based on an electrical characteristic model of voltage control, and the model schematic diagram is shown in FIG. Figure 5 Assume that the output voltage of the node is a constant value (between 0.6V and 0.8V), and use the equivalent resistance method to quickly obtain the current size of the SOFC stack node, as follows:
[0140] Assume that the output voltage of each node is a known constant U0 and satisfies the following equation:
[0141] V1=…=V j =U0;U0∈[0.6,0.8]
[0142] The current size of node j can be quickly obtained by the equivalent resistance method:
[0143]
[0144] 2. Simplification and linearization of the model
[0145] From the SOFC model established above, it can be seen that the stack has temperature state variables at node j and In addition, there are The four reaction concentration state variables are simplified by using the quasi-static assumption method to simplify the mass conservation sub-model. The quasi-static assumption assumes that due to the large heat capacity of the solid layer inside the SOFC cell, the thermal dynamic response time is long, and the time constant is in the order of hundreds of seconds; while the mass conservation and electrochemical reaction time constant inside the stack are in the order of milliseconds; therefore, the four reactant concentration state variables in the stack model are processed into steady-state processes, eliminating the four state variables and reducing the complexity of the model. According to the quasi-static assumption, the instantaneous change in the concentration of each substance is equal to zero, so the mathematical model of mass conservation is as follows:
[0146]
[0147]
[0148]
[0149]
[0150] After the above simplification, the battery stack model only has and Two state variables are used to convert the SOFC stack model into a state space model and abstract it into the following expression:
[0151]
[0152] y=g(x,u)
[0153] Among them, x and u are the state variable and input variable of the battery stack system, respectively, specifically:
[0154]
[0155] Different sub-models are mutually coupled, highly nonlinear, and require large and complex calculations. The present invention performs Taylor series expansion on the nonlinear stack model near the steady-state operating point (x0, u0) of the model to obtain its linearized model. The specific expression is as follows:
[0156]
[0157]
[0158] Among them, δx=x-x0, δu=u-u0, α(δx,δu) and β(δx,δu) are high-order small terms. After ignoring the high-order terms, the state space equation of the linearized nonlinear system of the battery stack in the neighborhood of (x0,u0) can be derived:
[0159]
[0160] δy=g(x,u)-g(x0,u0)=Cδx+Dδu.
[0161] 3. Observer Design
[0162] (1) Selection of observed variables
[0163] The stack temperature state observer accurately estimates the temperature distribution inside the SOFC stack through the limited measurable quantities u and y outside the stack. The simplified structural diagram of the stack and the temperature state observer is shown in the figure. Figure 6As shown. The selection of the observation variables of the observer is crucial. Different combinations of observation variables will affect the estimation accuracy and performance of the temperature observer. The combination of observation variables of the stack temperature observer is not unique. The selection of observation variables must ensure high accuracy while being low-cost and easy to measure. Therefore, it is of great significance to study how to select the observation variables of the temperature observer.
[0164] The first issue to be considered in the design of the stack temperature observer is the observability of the system. The observability of the stack system is determined by matrices A and C, that is, the observed system matrix (A, C) is completely observable. After the stack model is linearized near the steady-state operating point, the value of matrix A is determined after the state space model of the stack model is obtained. The value of matrix C is uncertain when the system output y is uncertain. Therefore, the value of the system output variable combination y determines the observability of the stack system. After satisfying the system observability, the output variable combination y is determined based on the evaluation index of the observation performance.
[0165] According to the previous modeling, the state variables of the system are The output of the system can be any combination of the 10 state variables. The observability of the system is determined by traversing all combinations of output variables. It is found that as long as Only one of these eight state variables needs to be output to ensure the observability of the battery stack system. There are many combinations of variables that meet the observability of the system. In order to achieve the design of low-cost, high-precision observers, it is necessary to evaluate the pros and cons of different combinations of variables. The condition number of the observability matrix is an important indicator. It is not only easy to calculate, but also can reflect the convergence speed and accuracy of the observer. The present invention uses a traversal method to obtain the condition number of all observable variable combinations. Considering that the observable state variables must have practical significance in actual engineering, the temperature sensor can be installed at a low cost for measurement. The fifth node is located at the end of the battery cell, and the temperature at its outlet is and It is easier to measure and its condition number is smaller, so the output y combination is selected and Since the Taylor series expansion is performed at the steady-state point, the input u of the stack must be determined at steady state. In summary, the observed variable combination is selected as the air inlet molar flow rate, the fuel inlet molar flow rate, and
[0166] (2) Structure of the stack temperature observer
[0167] By combining the selected observation variables, the air inlet molar flow rate, the fuel inlet molar flow rate, and The present invention designs a reduced-dimensional Romberg temperature observer to track and estimate the temperature inside the battery stack, and its specific form is as follows:
[0168]
[0169] in, is the estimated value of the temperature state variable, y is the output of the SOFC stack system, z is the intermediate state of the reduced-dimensional temperature observer, Q1 and Q2 are matrices of corresponding dimensions, and the specific steps are as follows:
[0170] S1: Eigenvalue group Compute the expected characteristic polynomial:
[0171]
[0172] In order to ensure the performance of the system, the expected eigenvalue group is taken as They correspond to the following expected values:
[0173] {-0.5339,0.5212,-0.5097,-0.4932,-0.0034,-0.0132,-0.0275,-0.0398}The coefficients of the expected characteristic polynomial are calculated as follows: They are 3.42e-09, 1.50e-06, 1.69e-4, 0.0071, 0.1193, 0.6823, 1.7629, 2.1420, 1 respectively;
[0174] S2: constitutes the (nq)×(nq) basic matrix F0:
[0175]
[0176] S3: Take any non-single matrix Calculate the inverse R -1 , calculate F = RF0R -1 , to ensure the non-singularity of matrix R, R takes the 8*8 identity matrix as shown below:
[0177]
[0178] S4: Take any constant matrix G is as follows, making {F, G} completely controllable, that is
[0179]
[0180]
[0181] S5: Solve the Sylvester equation TA-FT = GC for (nq) × n solution T;
[0182] S6: If q>1, go to the next step; if q=1, it constitutes P in the next step, skips the next step, and goes directly to S8;
[0183] S7: Judgment If it is not singular, go to the next step; if it is singular, go to S3;
[0184] Here, the observed variable is the temperature of the fifth node. and In order to ensure the observability of the system, C takes the following value:
[0185]
[0186] S8: Calculate H=TB;
[0187] S9: Calculate and divide Q1 is an n×q matrix, Q2 is an n×q matrix;
[0188] S10: In summary, the main form of the reduced-dimensional state observer is:
[0189]
[0190]
[0191] in, is the reconstructed state of the observed state, Figure 7 This is the structural diagram of the reduced-dimensional Romberg stack temperature state observer. Figure 8 It is the flow chart of the comprehensive algorithm of the temperature state observer of the reduced-dimensional Romberg stack.
[0192] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0193] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for estimating the internal temperature of a planar solid oxide fuel cell stack, characterized in that: The following steps are involved: Step 1: Construct a single cell mechanism model of the battery stack; Step 2: Use the quasi-static assumption method to simplify the mechanism model and obtain a simplified model; Step 3: Convert the simplified model into a state space model; Step 4: Perform Taylor series expansion on the state space model to obtain a linearized state space model; Step 5: Traverse the condition number of the observability matrix of all variable combinations in the linearized state space model to determine the observed variable combination; Step 6: Use the reduced-dimensional Romberg synthesis algorithm to construct a stack temperature observer based on the observed variable combination, collect the stack peripheral measurement data and input it into the stack temperature observer to obtain the estimated state of the stack internal temperature distribution; The process of building a stack temperature observer is as follows: Step 61: Select negative numbers as eigenvalues in the left half plane of the complex plane of the linearized state space model to form an eigenvalue group. Calculate the expected characteristic polynomial, the expression is: in, are the coefficients of the expected characteristic polynomial; s is a complex variable; n is the dimension of the coefficient matrix A in the linearized state space model; q is the rank of the coefficient matrix C in the linearized state space model; rankC = q; Step 62: The expected characteristic polynomials are formed into a (nq)×(nq) basic matrix F0, expressed as: Step 63: Select any non-single odd-number matrix in the basic matrix F0 Calculate the inverse R -1 , calculate the matrix F = RF0R -1 ; Step 64: Take any constant matrix in the basic matrix F0 Let {F,G} be completely controllable, the expression is: Where F is the matrix to be solved in the stack temperature observer; F n-q-1 It is the construction of the controllable matrix, which is used to determine whether {F, G} is completely controllable and whether the matrix G is selected correctly; Step 65: Solve the (nq)×n dimension matrix T of the Sylvester equation TA-FT=GC according to the coefficient matrix A and the coefficient matrix C, the matrix G, and the matrix F in the linearized state space model; Step 66: If q>1, go to step 67; if q=1, construct a non-singular matrix based on the coefficient matrix C and matrix T in the linearized state space model And proceed to step 68; Step 67: Determine if the matrix is non-singular If it is not singular, go to step 68; if it is singular, return to step 63; Step 68: Construct the matrix H based on the matrix T and the coefficient matrix B in the linearized state space model, expressed as: H = TB; Step 69: Calculate the matrix Q based on the non-singular matrix P and perform block processing. The expression is: Q1 is an n×q matrix, Q2 is an n×q matrix; Step 610: construct a reduced-dimensional Romberg stack temperature observer, the expression is: Wherein, y represents the output of the battery stack system; u represents the input of the battery stack system; is the estimated state of the temperature distribution inside the battery stack; z is the intermediate state quantity of the estimated temperature inside the battery stack, represents the differential of z; Step 611: Collect peripheral measurement data of the stack according to the observed variable combination, input the peripheral measurement data of the stack into the stack temperature observer, and obtain the estimated state of the temperature distribution inside the stack.
2. A method for estimating the internal temperature of a planar solid oxide fuel cell stack according to claim 1, characterized in that: The single cell sheets of the battery stack are divided into nodes according to the airflow direction, and the mechanism modeling is carried out according to the nodes. Each node mechanism model is composed of a mass conservation model, an energy conservation model and an electrical characteristic model; the energy conservation model includes a solid layer temperature dynamic model and an air layer temperature dynamic model.
3. A method for estimating the internal temperature of a planar solid oxide fuel cell stack according to claim 2, characterized in that: The quasi-static assumption method is used to reduce the number of state variables in the node and obtain a simplified model; the simplified model includes two state variables.
4. A method for estimating the internal temperature of a planar solid oxide fuel cell stack according to claim 1, characterized in that: A Taylor series expansion is performed at the steady-state operating point of the state-space model.
5. A method for estimating the internal temperature of a planar solid oxide fuel cell stack according to claim 1, characterized in that: The determined combination of observed variables includes input air flow rate, fuel gas flow rate and stack outlet terminal temperature.
